Accuracy in Using Dual-Isotope Schilling Test to Measure Urine Samples

Document Sample
Accuracy in Using Dual-Isotope Schilling Test to Measure Urine Samples Powered By Docstoc
					Accuracy in Using Dual-Isotope Schilling Test
to Measure Urine Samples: A Multicenter Study
Borys R. Krynyckyi and Lionel S. Zuckier

Department ofNuclear      Medicine, A1be,@Einstein College ofMedicine,        BrotvçNew York

                                                                      performed where the radiotracer is supplemented with cx
As a component of our quality assurance program, this multi           ogenous intrinsic factor (IF) in an attempt to correct the
center study was performedto characterize the magnitude and           abnormality.
types of error present in measurement of typical dual-isotope            To improve convenience and reliabilityof the examina
Schillingtest (DIST)urine samples. Methods: A panel of three          tion, a dual-isotopevariationwas conceived by Katz et al.
simulated DISTunne samples was formulatedcorrespondingto              in 1963, which performed both stages of the test simulta
diagnoses of normal excretion, malabsorption and pernicious  neously (7). The commercial version of the dual-isotope
anemia and was distributedto sight hospitals in our regional Schilling test (DIST), based on modifications reported by
area (three noviceand fiveexperienced users). Count-ratedata
                                                             Bell et al. (8), utilizes two 0.25-pg capsules ofvitamin B12,
and unne volume measurements from each site were analyzed
for accuracy against the predictedvalues and a carefullymea                              of
                                                             one labeledwith 0.8 j@Ci 58Coand lackingexogenous IF,
suredgold standardand were correlated   withthe methodology                                          and
                                                             the other labeled with 0.5 1uCi @@Co preincubatedwith
and equipment used. Results: Three of 24 results were unin   IF-containinghumangastric secretions. Matched @@Co        and
terpretable due to an overly low ratio of intrinsic factor bound to   58@   standards,   each   containing   2%   of the   respective    ad
free vitamin B12 excretion (B/F ratio), inconsistent with possible    ministered doses per milliliter, are included in the kit for
diagnoses. In 20 of 21 interpretable samples, results corre           reference purposes. The urine sample is identical for both
sponded to the appropnate diagnoses, w@itypicalvalues noted           phases of the examination and the ratio of IF-bound-to
in 18 ofthe cases and slightly atypical yet diagnostic values seen    unbound (free) vitamin B12 excretion (B/F ratio) may be
in the remainingtwo cases. In only one sample did values              used to evaluate the corrective effect of IF, even when
correspond to an erroneous diagnosis Qow normal or partial            urine collection is incomplete (9).
rnalabsorption rather than pernicious anemia). The four major
                                                                        This investigation was initiated as part of an ongoing
discrepancies (test failureor misdiagnosis) were largelyattiibut
able to blunders and were limitedto two of the three novice           quality assurance program (10). We have occasionally
sites and to a single experienced site which had grossly maccu        noted irregularresults of the DIST at our affiliatedhospi
rate raw data (background greater than sample counts). tals, some of which relate to biological issues involved in
ConclusIon:Quantitationfvitamin excretioninDISTurine coadministration of the two phases of the examination
                            o         B12
samples is a reliablemethod of evaluationwhen performed by (11 ), while others have been attributed to various technical
reasonably expenenced and competent dinical laboratones. Im     faults in performance of the assay, including stochastic
proved accuracy may be obtalned by increasingthe stochaslic variation due to poor counting statistics (12,13). To char
certainty of the count data and by more careful measurement of acterize the magnitude and types of error present in the
the sampleand urinevolumes.                                     assay of the two labeled vitamin B12 moieties within DIST
Key words: Schilling test; quality assurance; dual-isotope stud samples, we created a panel of three calibratedmock urine
ies; urinemeasurement                                           samples which were distributedto two ofour own affiliated
                                                                      hospitals in addition to another six participating centers in
J NucIMed1995;36:1659-1665
                                                                      our regional area. Biological issues involved in coadminis
                                                                      tration of the two vitamin B12 capsules have been previ
                                                                      ously reviewed (11) and were not relevant to this artificially
                                                                      constituted panel.
    he Schilling test, devised in 1953 (1,2), remains the
most practical and effective method of evaluating vitamin             MATERIALSAND METhODS
B12 absorption following an oral test dose of radioactive
cyanocobalamin     (3—6). When urinary    excretion   of the la       Five two-patient DIST kits were used (Dicopac, Amersham
beled vitamin is low, a second stage of the test may be
                                                                      HealthCare,Arlington, eights,IL). The contentsof six @Co
                                                                      andseven 58@ vitaminB12 capsules were each dissolved into 100
                                                                                         salineandcountedon a gammascintilla
                                                                      ml phosphate-buffered
                                                                      tion counter. They were furtherdilutedwith water to attaincount
  Received                 a                                          rateconcentrationscorrespondingto the manufacturer's2%stan
   Forcorrespondenceor reprintscontact UonelS. Zudder,Departmentof Nu
clearMed@sne,Aibe@t nst@n
                   @ CdlegeofMedkine,       1300Morris arkAve.,
                                                       P       UNman dardsat 2 wk priorto kit expiration,i.e., the conditionswe wished
121,Bronx,V 10461.
          N                                                           to simulate. These preparations,tinted appropriatelywith yellow

Accuracy of Schillin9 Test Measurements •
                                          Krynyckiand Zuckier                                                                           1659
                                                         TABLE I
                                           Designand Formulationf MockUnneSamples

      anemiaFree  diagnosisNormalMaisbsorptionPernicious
      .424.205.11IF-Bound excretion (%)21
           seCo-Vitamin B12
                 57Co.Vitamin B12 excretion (%)21
              2% 58Co stock added
              2% 57Co stock added
              water added (ml)12,785.7412,009.6912,452.49

and blue food coloringwere thereafter used as stock solutions RESULTS
fromwhich the mock urine samples were constituted and as new
                                                                The equipment and methodology used at the eight sites
2% standards distributedfor referencepurposes.
                                                              are shown in Table 3. Three of the laboratories (sites 1—3)
    Total urine volumes and values of total 57Co and 58Co
vitamin  B12 urinary excretion, typical for subjects with nor
                                                              had not previously performed DIST and therefore repre
mal absorption, malabsorption and pernicious anemia (urine        sent a group of novice users in contrast to the remaining
samples A, B and C, respectively) were selected (Table 1). five sites (sites 4—8)   which were using this examination and
Calculatedamountsof the new 57Coand 58Co2%standard/              therefore constitute a relatively experienced user group.
stock solutions were added into appropriate volumes of tap Raw data for the eight sites appear in Appendix 1 and
water, as measured by weighing the constituent components calculatedvalues and apparentdiagnoses appearin Appen
on sensitive electronic balances, to enable preparation of eight dix 2.
replicate samples (Table 1). Trace amounts of yellow food Total Urinary Excretion
coloring were added to simulate the appearance of urine. The       The free B12 excretion and B/F ratios for the 24 points
estimated final concentrations of isotopes were accurate to
                                                                 appear in Figure 1, along with the usual range of values
within 0.1% of the intended values. Individual urine samples
                                                                 encountered in various disease states (15). Three samples
were subdivided by weight into either of two types of 24-hr
urine collection containers to an accuracy of within 0.01% of
                                                                 had a B/F ratio less than 0.6 (Bi, B2 and B8), which does
the intended volume.                                             not correspond to known disease states but suggests in
  Permission of the chief-of-service at each hospital was ob      stead an erroneous result (16). In 20 of the remaining 21
tamed prior to initiation of the studies, and the nature of the   samples, the results corresponded to the appropriatediag
exercise was not disclosed to the technologists performing the    noses. Eighteen results were typical of the corresponding
procedure. Approximately 1 wk after constitution, each of the     diagnoses whereas 2 results (A8, B6) were slightly outside
eight study sites received three filled urine collection jars,    the usual range but still suggestive of the correct disorder.
marked with fictitious names, the aliquots of 2%57Coand 58Co      Only one sample (C2) would have been misinterpreted
standard solutions and copies of the worksheet accompanying       based on the urinemeasurements, indicatinglow-normalor
the commercial kit. After each hospital's measurements and        partial malabsorption rather than IF-deficiency, although
calculations were completed and returned to us, information       cautious interpretationmay still have raised this possibil
regarding the equipment and methodology used was solicited        ity, based on a B/F ratio approaching 1.2.
from the technologists who performed the examination.                Of the four significant discrepancies discussed above
   Based on data returned from each center, the percent per
milliliter (%/ml)and total @@Co 58Coexcretions were cal
culated and used to determine the B/F ratio, according to a                               TABLE 2
customized spreadsheet we developed (Quattro Pro, Borland                      Mean Values and Usual Range (15)
International, Scotts Valley, CA). Standard errors of the esti
mate were also determined based on stochastic considerations
(14). Unused samples of the mock urines and 2% standards                 excretionB/FDiagnosis excretionB12
were retrieved from center 8 and 5 ml volumes were carefully                                    (%)(%)ratio
alliquoted and counted for 167 mm each to obtain our own
measured values, which were generated with a high degree of        Normal20               (9-33)20
statistical validity. Measured values were compared to pre
                                                                               syndromes4        (0-8)4        (0-8)0.6-1.1
dicted isotope excretions calculated based on amounts of the
2% per milliliter standards added to each of the samples.
Interpretation criteria were based on revised guidelines in               patients
                                                                     *@fl@e                   a
                                                                                 withpernicious nemiagiveratiosinthe rangeof 1.2
cluded in the commercial product information sheet (Table 2).     to 1.4.

1660                                                                                        Vol.   No.   September 1995
                                                            The Journal of Nudear Medicine • 36 • 9 •
@                             —@

                                     I                                                                       0   Urine A
             3.6@—@                7
                                                                                                             0 UrineB
             2.2-1'             4@                                                                           L,@I Urine   C

             2.0-I               L:@i+                                                                                 V
                                                                                                             + Predicted alue
             1.8 - I
                         IL@<                                                                                        Value
                                                                                                             x Measured
@                        I
             @1.6-I                      L@L@1        \\
       .@                I
        a)               I      @-DEFiciENCY'@j
         @3I 4       -
       w                                         1'
             12      -
                                                           r —                  á@r@l
                                                           I                          LJL,c+       0
       V                                                   I                                             0                          I

@      0@    08 -                        0                     \                     NORMAL                                         I
       co                MALABSORPTION           I
             0.6                                                                                                                        o8
             0.4 —
             0.2 —
@                    @8                                     i         I      Q1 I         I                                I
                  vo@                    5                 10        15        20        25                               30                 35
                                                            Free 58CoExcretion PatientDose)
    FiGURE1. FreesaCoexcretion                      samplestested.Predicted alues,basedontheamount fstocksolution
                                       forthe24 urine                     v                      o
    added and measured values, obtained by prolonged countingofsamplesfrom site 8, are also displayed. Ranges typicalof normal excretion,
    maiabsorpbon and pernicious anemia, as suggested by the manufacturer, are iflustratedby the dashed boxes. Values lyingoutside of their
    expected ranges have site numbers indicatedto aid InIdentification.Because of its negative value, urine sample B, site 8, has been replotted
    at the origin.

    (test failure or misdiagnosis), three occurred at two of the           volume in graduated cylinders. Two of these sites (3, 4)
    novice sites, while the fourth occurred at site 8 which had            were accurate to within several milliliters, while site 5
    grossly flawed raw counting data (background counts were                                                                ml
                                                                           consistently overestimated urinevolume by 77—79 (Fig.
    greater than urine counts, see Appendix 1). The two minor              2). This inaccuracy may be related to a systematic error
    discrepancies (correct diagnosis with atypical values) cc              arising from measurement       of urine volumes     greater   than 1
    curred at sites 6 and 8, both experienced locations. For the           literusing a 1000-migraduatedcylinder, althougha definite
    competent sites (excluding the novice users and site 8,                cause cannot be ascertained.
    which had obviously flawed data) average errors (mean ±
                                                                           Percent Vitamin B12 Excretion per Milliliterand
    1 s.d.) were 0.84% ±0.74% for             excretion and 1.12%
                                                                           Stochastic Error of Measurement
    ±1.07% for 58Co excretion.
                                                                              The low counting rates seen with 58Co resulted in a
       To furtherunderstandand characterize the cause of the
@                                                                          greaterstandarderrorof the measurement than with
    observed discrepancies, the total excretion of each isotope
                                                                           and, to a large degree, much of the discrepancy between
    was broken down into the independentconstituent compo
                                                                           observed and predicted values can be attributed to this
    nents of total urine volume and percent excretion of each
                                                                           statistical variation. In general, the observed values distrib
    isotope per milliliter.
                                                                           uted around the predicted percent excretion, with the sin
    Urine VoIum                                                                                                            i
                                                                           gle exception being measurement of @Con urine C, in
       Five of the sites (1, 2, 6, 7, 8) measured urine volume by          which observed values and our own measurement were
    sighting the level relative to volume graduations stamped              slightly below predicted values (Fig. 2). Furtherinvestiga
    on the urine-collection container walls and they overesti              tion of this phenomenon led to analysis of a small amount
    mated urinaryvolume by an average of 117 ± ml (mean                   of black sediment present on the bottom of the urine C
    ±1 s.d.) (Fig. 2). The remaining three sites measured urine           carboy which was counted and found to contain @@Co          ra

    Accuracy of Schilling Test Measurements •
                                              Krynyckiand Zuckier                                                                            1661
@                                                                                                                                        1@

                                 URINEA                        URINE B                             URINEC
@                                           I
@         0.015 +
                    .                           I       •

@              0    .
          @0.005.   .+
         11111111.oo1c                                      IIIIItIL_+i@ihl@1w
           .             I     2345678                  1     2345678

                                                                                                                                                       I Co-57
          o.02e1234@6780.020                                                                                                             @4OO          Dco-58

                                                             x Measured
          0.005                                                                                                   *
@                                                   +             •J.I@                                     .
               0    .+                                  .1


          11111111@0011                                 Illitill+1.1.11
@          .                                                                                                                             0.9
                                                                                                                                    8 0.8 FRACTIONtnH0nUf

           175C1                                                                                                                    S    0.7
                                                                   Predicted                                                        C.   0.6
                             .              .
                                                                                                                                    2 0.5
           1650          .
                                                                                                                                    g 0.4
@          1600     -                                         S                              •      I     ••
@                                 •.                                                                                                   0.2
           1550                                         I                      •   I
                                                                                                                                    t@) °.i
           1500     .                                                                    +         ••
                                                                                                                                    C)                    2     3    4          5      6   7       8
           1450     .                               +                 S                                                                                                  SITE

                         11111111+                      11111111.                            11111111
                         I 2345678                      1 23 45678                           1234S678                      FIGURE 3. Netcountrates (top)and backgroundrates (m@dle)
                                                                    5ITE                                                           and
                                                                                                                          for @Co eaCo 2% standards. At each site, 1 ml standard was
                                                                                                                          dilutedintothe measuredvolume,withthe exceptionof site 1 where
                    urine           andisotopeexcre
    FIGURE2. Measured volumes(bottom)                                                                                      6 ml were diluted into I liter of volume. Because sample counting
                      o                            (middle).Errorbars
    bone per milliliter f urineforeaCo (top)and @@Co                                                                       was performed on different dates with differinggeometries, each
    represent 2 s.e.e. based on counting uncertainty.                                                                      count rate sensftMty reflects the particular conditions under which
                                                                                                                           thesamplewascounted                        c     the
                                                                                                                           sensitivityof one instrument to another.
    dioactivity. The carboys containing the remainder of urine
    samples A and B were inspected and did not contain sed
    iment, and our own measurements of these samples, on                                                                   DISCUSSION
    samples obtained from site 8, did not deviate appreciably                                                                 When measuring DIST urine samples, diagnostic error
    from predicted values (Fig. 2), indicating that this was an                                                            can be attributed to inaccurate measurement of the total
    isolated phenomenon.                                                                                                   urine volume, statistical uncertainty due to low count 1ev

                                                                                                                      TABLE 3
                                                                                                     Equipmentand Methodology
                                          measurementCrystal                                                                       countingVolume
      sample1NovIce3―                                      Analyzer ChangerTrne (mm)Volume (ml)                                                DuplicateTotal                     urineCounting
      GC2NovIce2―WELL END         SCA  Manual51000                                                                                                NoSpecimen                        bottie1000       ml
      bottleMicropipette3Novice3―THRU Manual105                                                                                                   NoSpecimen
      GCMicropipette4Experienced2―WELL Automated205                                                                                          Standard, Urine1000          ml
      GCMicropipette5Experienced3―WELL Manual104                                                                                                  No2000                  ml
                                    DCA                                              Automated205                                                   No1000                  ml
                                S@A Manual104                                                                                                       NoSpecimen
                                   DCA    Automated102                                                                                        UrineSpecimen
                               SCA                                                   Automated303                                                  UrineSpecimen

                                         SCA=slngle-channel analyzer,                                 DCA=dual-channel          analyzer.

    1662                                                                                                                                               Vol.    No.  September 1995
                                                                                                                      The Journal of Nuclear Medicine • 36 • 9 •
                                                          APPENDIX 1
                                                           Raw Data
   C157Co                         (mm)Background@@Co            Standard@Co       standardUrine       AUrine         BUrine
                 CPM                                                              129
           @COCPM                 5494          285026          —1             11315923         1067        1123            282
   1600257COCPMvolume(ml)5                                                                        16501333    15502628

          58CoCPM                10207         24912484         —1281          2738644           322         262            276
   1700357Co                                                                                      I 700242    1600382
            @COCPM               2075           8113913/13372   —1661/1705   5457/5636565/564253/252       113/119      118/119
   1510457COCPM                                                                                   1570164/1671470278/280

          5800   CPM             10176         22717309         —1881          4546705           340         259            257
   1510557Co                                                                                      1560288     1460381
          58CoCPM                2014           4314456         —551           4741507           190          72             71
   1585657COCPM                                                                                   1640109     1530209

         58CoCPM                 1044          10315770         —840           4232512           220         127            135
         Urine olume
   1625757COCPM (ml)10                                                                            1675151     1550251

          58CoCPM                1064           3417369         —1512          4068320/326 92/100          45/45        48/42
          Urine   (ml)10
   1600857Co volume                                                                               1680114/109 1580168/173
                 CPM                                            5999
          5800   CPM             30224/254 206/2291             —1974          3783588/552 335/310    191/186     233/211
          Urine volume (ml)30                                                                 1700196/196 1550322/308 1600

els and miscellaneous causes, such as inaccuratepipetting,      erroneous because of B/F ratios below 0.6, which would
incorrect window placement, cross-contamination of sam          have necessitated repetition. In only 1 of the 24 samples
ples or other blunders. Whereas the Dicopac Users Guide         would an incorrect diagnosis have been made (Fig. 1).
(16) suggests a protocol for calibrating and checking en        What is remarkable is that significant errors only occurred
ergy windows on gamma counters using a “mock      urine―    at two of the three novice hospitals, and at site 8, which
sample, the purpose of the simulatedurines in this exercise     had grossly inaccurate count rate data, possibly due to
was to evaluate all aspects of DIST measurement.                invalidcountingwindows or contaminationof the counter.
   The urine samples analyzed in this exercise were de          Withthe exception of site 8, it appearsthatthe experienced
signed to represent a worse case scenario. To achieve this      centers performedthe test in a generally accurate manner,
end, the samples were submitted in a routine manner, so         resulting in data which would have led to appropriatedi
that the technologists performing the examination were          agnoses. A blunder in recording or measuring one of the
unawareof the natureof the study. The kit was distributed       urine samples is the most plausible explanation               for the
1 wk prior to its simulated expiration, and the urine vol       error at site 1. The cause of errors at site 2 cannot be
umes in which the activity was diluted were in the upper        ascertained definitely, although the stochastic error was
range of normal. These factors combined to yield low ra         largest at this site. The minor error seen in the urine B
diocobalt concentrations with resultant poor counting sta       sample from site 6 is possibly related to their imprecise
tistics. Nonetheless, excretion values were chosen in the       method of measuring standardvolumes with a disposable
midrangeof typical values, which tended to increase diag        syringe.
nostic accuracy because of the large changes necessary to          Stochastic uncertaintydue to poor counting statistics, as
shift values from the midpoint of a given diagnostic cate       demonstrated by error bars in Figure 2, is a widespread,
gory into another.                                              yet easily rectified, source of error which can be reduced
   Actually, quantitation of the cobalt isotopes in the         by either increasing the volume of the samples counted or
Schilling samples was quite accurate, resulting in values       by lengthening the counting time. Counting 2 or 3 ml of
thatwould have permittedcorrect diagnoses to be made in         urine for only 10 or 20 min, as was performed in some of
20 of the 21 interpretablecases (95%frequency). Typical         the sites, is inadequate at the low isotope concentrations
values were seen in 18 of these samples and atypical,           prevailing during the DIST. The product insert for the
although still diagnostic, results were observed in another     commercial dual-isotope kit (15) recommends counting
2. In 3 of the original 24 cases, results were identified as         largestpracticalvolume― at least 10 ml, while the
                                                                “the                          and

Accuracy of Schilling Test Measurements •
                                          Krynyckiand Zuckier                                                                  1663
                     APPENDIX2                                                       APPENDIX2 (contInued)
 Percent ExcretionCobalt Isotopes and Resuftant Diagnosis
(normal)ApparentSiteIsotopeValue sample A                                                        sampleC (IF

 diagnosisI           ±
                       s.d.      B/F ratio                         SiteIsotopeUrine ±.d.B/F ratioApparent

                                1.13±0.03    Normal                deficiency257Co9.00±0.4858Co7.54±1.981
                                                                                              .90 ±
        58Co    18.64±0.37
 2      57Co           0.55
                22.43 ±
                                1.12±0.12 Normal                                                       0.32Low
                                                                                                   .19 ±            normal/Mild
        58Co    19.94±2.07                                         malabsorptlon357Co8.59±0.1458Cc@4.14
 3      57Co           0.20
                20.32 ±
                                1.03 ±       Normal
        58Co           0.37
                19.71 ±                                                                          0.15IF
 4      57Co    22.05±0.45                                                           ±0.292.07
                                I.08 ±       Normal
        58Co 20.37±1.36                                            deficiency557Co8.42±0.1558Co3.85±0.322.19±0.19lFdeflciency

 5                  0.24
         @@Co21.63 ±
                                1.05 ±       Normal
        58Co           0.48
                20.55 ±
 6      57Co           0.41
                23.74 ±
                                I .00 ±      Normal
        58Co    23.63±1.16                                         deficiency757Co9.34                 0.26IF
                                                                                                   .66 ±
 7      57Co    22.99±0.48
                                0.90 ±
                                      0.05    Normal                      0.3758Co4.28 ±
       68Co     25.51±1.20
 867Co 57Co     20.13±0.29                                         ±0.952.18±0.49IFdeficienCy857Co4.89±0.2258Co1.35±0.813.64±2.19IFdefIcIenCy
                                0.60 ±       Atypical normal
              33.38 ±.96Urine
                              sample B (malabsorptlon)
 diagnosisI            s
                      ±.d.      B/Fratio
                                                      associated Users Guide (16) bases its examples on count
                                                      ing of 500-ml urine volumes. Other than site 1, which
                          0.05 ±    Nondiagnostlc    utilized a specialized counting method, none of the labora
        58Co 18.47±0.36
                                                      tories measured more than 5 ml of urine and, practically
 2       @@Co 1.51 ±0.38                             speaking, typical clinics do not have access to gamma
                          0.45 ±    Nondlagnostic
        58Co  3.32±1.64                              counters with larger capacities. The product insert (15)
 3      57Co  3.48±0.13
                                                      also recommends countingfor “an    appropriatetime,―gen
                          0.92 ±
                                0.07 Malabsorption                                 20        O
                                                      erally specified as “usually min.― nly three ofthe eight
        58Co  3.77 ±0.28                             sites met or exceeded this recommendation. Counting for
 4      57Co  4.24 ±0.31                             prolonged periods of time is possible when the gamma
                          0.79 ±
                                0.18 Malabsorptlon    counter is equippedwith an automaticsample changer and
        58Co  5.38±1.18                              a multichannel analyzer that can be run overnight; with a
 5      57Co  3.88±0.11                              single-well counter, it is impractical to occupy a technolo
                          1.02 ±    Malabsorptlon    gist and use the device for several hours.
        58Co  3.79±0.31
                                                         In several cases, error in the calculated %/mlexcretion
 6      57Co  5.06±0.22                              of cobalt isotopes was greater than that due to statistical
                          1.16±0.24 AtypIcal
                                                  @on variation alone (Fig. 2). This suggests that a component of
        58Co  4.37 ±0.90                             the observed errororiginatedfrom another source, such as
 7      57Co  3.95±0.33                              imprecision in dilution of the standards, placement of win
                                0.96 ±
                                      0.27    Malabsorption       dows or in pipetting of the counting samples. While count
       58Co      4.11±0.93                                       ing of duplicate samples is advocated in the product insert
 857Co 57Co     —1.89±0.20                                     (15), this was only performed on samples at three sites and
                                0.23 ±       Nondiagnostlc       on standards at one site (Table 3). Evaluation of these
               —8.40 0.75
        56Co0.86±0.12±                                           duplicate samples confirms that the differences between
                                                                  values was generally greater than that originating from
                                                                  counting errors alone (Appendix 1). At present, many nu

1664                                                                                     •       No. September1995
                                                              The Journalof NudearMedicine Vol.36 • 9 •
clear medicine technologists do not perform in vitro work       niques: We suggest counting duplicate samples to monitor
on a routine basis and may lack accurate pipetting and          precision of measurements. Mock urine samples provide
other measuring skills. Poor methodology may also be a          an excellent method of monitoring all technical aspects
factor, as site 6 was aliquoting volumes with disposable        involved in this test, from measurement of urine volume to
plastic syringes, which is an inaccurate method. Careful        calculation of percent excretions and BIF ratios, and may
attention to accurate volume measurement, including use         provide a means of overcoming the startup difficultiesen
of increasingly available semiautomated electronic pi           countered by novice and inexperienced users.
pettes, may help rectify this component of error.
   Count rate sensitivity for the two isotopes, in additionto   ACKNOWLEDGMENTS
the spill-down fraction, were calculated at each site based        This study is dedicated to the memory of L. Rao Chervu, who
on count rates of the standard samples (Fig. 3) With the                                    in
                                                                wouldhaveenjoyedparticipating thisprojectandwhose par
exception of site 1, which used a counter with unique           ticipation was sorely missed. Appreciation is extended to Mcdi
geometry, count rate sensitivities were similar at the van      PhysicsInc., AmershamHealthCare for supplyingthe radiophar
ous sites. Although wherever possible, counters with a          maceutical kits and to the following sites for participation in the
3-inch diameter crystal are preferableto 2-inch crystals, to    study (alphabetical and not study order): Beth Israel Medical
allow improvedefficiency in detecting the energetic gamma                                         o
                                                                Center,New York,NY; BronxDepartment f VeteransAffairs
emissions of 58@ (13), no differences were noted between        Hospital, Bronx, NY; Lenox Hill Medical Center, New York,
counters with 2- or 3-inch ciystals in our limited sampling.    NY; Montefiore Medical Center, Bronx, NY; New York Hospital
                                                                Medical Center, New York, NY; St. Luke's Hospital, New York,
Because of differences in models of counters and methods
                                                                NY; St. Vincent's Hospital, New York, NY; Weiler Hospital of
of energy calibration, it was impractical to analyze the                                           B
                                                                theAlbertEinsteinCollegeof Medicine, ronx,NY. Dr.Zuckier
appropriateness of the specific @@Co 58Q@         windows.      is partially supported by a National Institute of Health Physician
Spill-down fractions varied between 0.1 and 0.6 at the          Scientist Award (1K11 CA01503).
different institutions (Fig. 3), probably because of vana
tions in the energy resolution of the detectors and the         REFERENCES
particularenergy windows chosen.                                 1. SchilhingRF. A new test for intrinsic factor activity. I Lab Clin Med
  As a general rule, sites that measured urine volume by            1953;42:946—947.
graduations on the collection bottles overestimated urine                                factorstudiesII. The effectof gastricjuiceon the
                                                                 2. SchillingRF. Intrinsic
                                                                    urinaiyexcretion ofradioactivity after the oral administration of radioactive
volume by over 100 ml, which in turn inflated the final             vitamin B12. I Lab Clin Med 1953;42:860-866.
urinary excretion of the cobalt isotopes by 6%—7% cxof         3. Bnedis D, Mcintyre PA, Judisch J, Wagner HN, Jr. An evaluation of a
pected values (Fig. 2). It was not possible to retroactively       dual-isotopemethodforthe measurementofvitaminB12absorption.JNuc/
                                                                   Med 1973;14:135—141.
identify differences between the brands of collection bot       4. Nickoloff EL. Alternatives of vitamin B12 radioassays: the Schilling test.
tles but subsequent comparison of the bottles revealed             Ligand Q 1979;2:27—29.
errors of 89.8 ± and 89.5 ± ml (mean ± s.d.) when
                 4.6           5.3              1               5. DomstadPA, Choy YC, Kim EE, DeLand FH. Reliabilityof the dual
                                                                   isotopeSchillingtest for the diagnosisof perniciousanemiaor malabsorp
measuring a 1508-ml sample. Measurement of the urine               tion syndrome.AmI ClinPath 1981;75:723-726.
volume in a graduated cylinder represents a simple and          6. Fairbanks VF, Wahner HW, Phylisky RL. Tests for pernicious anemia: the
reliable alternative.                                                       test―.
                                                                   “Schilling Mayo ClinPivc 1983;58:541-544.
                                                                                                                a             o
                                                                7. Katz JH, DiMaseJ, DonaldsonRM Jr. Simultaneous dministrationf
                                                                   gastric juice-bound and free radioactive cyanocobalamin: rapid procedure
CONCLUSION                                                                           betweenintrinsicfactordeficiencyandothercausesof
                                                                   for differentiating
                                                                   vitamin B12 malabsorption. I Lab Clin Med 1%3;61:266-271.
   Quantitationof vitamin B12 excretion in DIST samples                         JM,NelsonMG.Simultaneous
                                                                8. BellTl@Bridges                      freeandboundradioactive
has proven to be a reliable method in a variety of regional        vitamin B12 urinary excretion test. I Clin Path 1965;18:611—613.
laboratories when performed by experienced practioners.         9. ChanarinI, WatersDAW.FailedSchiuingtests. ScandlHaemat 1974;12:
There appears to be difficulty initiatingthis procedure for     10. LamkiLM, HaynieTP, PodoloffDA, Kim EE. Qualityassurancein a
the firsttime, which is probablyrelatedto placement of the         nuclear medicine department. Radiology 1990;177:609—614.
scintillation counter window settings and inexperience                                            evaluationofpernicious anemia:current
                                                                11. luckier LS, ChervuLR. Schilhing
                                                                    status. JNucl Med 1984;25:1032—1039.
with the protocol. Urine volume must be measured by a           12. Smith JP, Graham MM. Schilhingevaluation of pernicious anemia [Letterj.
reliable method, such as use of a graduated cylinder, be           I NuciMed 1985;26:1099-1100.
cause volume markings on the collection bottles were            13. ZuckierL5, ChervuLR. Schillingevaluationof perniciousanemia[ReplyJ.
                                                                   JNuclMed 1985;26:1100.
highly inaccurate. A widespread source of error in mea                      Al,                            s
                                                                14. Sorenson PhelpsME.Nuclearcountingtatistics.In:Physics    innuclear
surement of the cobalt isotopes stems from poor counting           medicine. New York: Giune & Stratton; 1980:100-118.
statistics. This can be improved by increasing the sample       15. Medi+PhysicsInc. Dicopac Product Information.ArlingtonHeights, IL:
                                                                   Medi-Physics, Inc. 1990.
volume and counting time. Additional errors, such as pi         16. Medi+PhysicsInc. DicopacUsers Guide.ArlingtonHeights,IL: Mcdi
petting and dilution, can be avoided by using proper tech           Physics,Inc. 1990.

Accuracy of Schilling Test Measurements •
                                          Krynyckiand Zuckier                                                                            1665